Real Estate MCP Server for AI agents — query Zillow, Redfin, Rightmove (UK), Singapore HDB, Boliga (Denmark), and apartments inventory through a single Model Context Protocol endpoint. Pay-per-call on Apify at $0.024 per result vs $299-2,999/month for RealEstateAPI, $59-549/month for Rentcast, and $10K+/year for HouseCanary or ATTOM Data.
Real-estate data is fragmented by source, jurisdiction, and update cadence. Zillow has the largest US listing inventory; Redfin has the best comparable-sales feed; Rightmove dominates UK property; Singapore HDB and URA publish authoritative resale-flat and private-property transaction prices; Boliga is the Denmark standard. Each of these used to require a separate scraper, a separate auth scheme, and a separate parser — which is why proptech engineers, family-office analysts, and AI agent builders have historically spent more time on plumbing than on analysis.
The Real Estate MCP Server consolidates these sources behind one Model Context Protocol endpoint that Cursor, Claude Code, Claude Desktop, and Replit Agents can call natively. Ask the agent for “active 3-bedroom listings in Austin under $700K with Zestimates” or “all HDB resale flats in Bishan in the last 90 days” and the MCP server resolves the right tool, runs it pay-per-result on Apify, and streams structured rows back to your agent context. 2,174 lifetime runs and counting — the most-used MCP server in the NexGenData fleet.
Sample output — Zillow active listings via MCP
| address | price | beds | baths | sqft | zestimate | days_on_market |
|---|---|---|---|---|---|---|
| 1342 Oak St, Austin TX 78704 | $685,000 | 3 | 2 | 1,820 | $702,400 | 14 |
| 208 Hillcrest Dr, Austin TX 78704 | $649,000 | 3 | 2 | 1,640 | $631,800 | 22 |
| 4711 Bouldin Ave, Austin TX 78704 | $729,500 | 4 | 2.5 | 2,140 | $745,600 | 7 |
| 891 Travis Hts, Austin TX 78704 | $598,000 | 3 | 2 | 1,580 | $612,200 | 31 |
Same query schema works for Redfin (comparable sales), Rightmove (UK postcodes), HDB (Singapore block/town), and Boliga (Denmark kommune). The MCP server normalizes field names across providers so your agent gets price, beds, baths, sqft, address on every row regardless of source.
Why this beats HouseCanary, ATTOM, CoreLogic, Rentcast & RealEstateAPI
| Feature | NexGenData Real Estate MCP | HouseCanary | ATTOM Data | Rentcast | RealEstateAPI | Zillow Zestimate API |
|---|---|---|---|---|---|---|
| Starter cost (10K records / mo) | ~$24 (pay per result) | $10K-50K/yr min | Enterprise quote only | $59-549/mo | $299-2,999/mo | Deprecated for new devs |
| Pricing model | Per-result, no commitment | Annual enterprise contract | Annual enterprise contract | Monthly subscription tiers | Monthly subscription tiers | Closed to new partners |
| Geographic coverage | US + UK + Singapore + Denmark | US only | US only | US only | US only | US only |
| MCP-native (AI agent ready) | Yes — Claude / Cursor / Replit | No (REST only) | No (REST only) | No (REST only) | No (REST only) | n/a |
| Zestimate + Redfin + Rightmove in one call | Yes | HouseCanary AVM only | ATTOM AVM only | Rentcast estimate only | RealEstateAPI estimate | Zestimate only |
| Singapore HDB + URA transactions | Yes (official datasets) | No | No | No | No | No |
| API key required | Apify token (free tier) | Enterprise contract | Enterprise contract | Stripe + signup | Stripe + signup | Closed program |
| Output formats | JSON / CSV / Excel / MCP-stream | JSON / CSV | JSON / CSV | JSON | JSON | n/a |
| Time to first row | ~30 seconds | 2-6 weeks (procurement) | 2-6 weeks (procurement) | 5 minutes (signup) | 5 minutes (signup) | n/a |
Use cases
AI agent: build a “Zillow chatbot” in 20 minutes
Connect Claude Desktop or Cursor to the Real Estate MCP server, give it a system prompt with your buy-box criteria, and the agent will pull active listings, comparable sales, and Zestimates on demand. No glue code, no separate scrapers — the agent calls search_listings, get_comparables, and get_zestimate as tools. This is the fastest path from “I want a real-estate AI app” to “shipped MVP” — most indie hackers we talk to have a working prototype within a single Saturday session.
Family office: cross-jurisdiction yield analysis
A family office tracking residential yield across US (Austin, Miami), UK (London, Manchester), and Singapore (HDB resale + URA private) used to need three separate data subscriptions costing $40K+/year combined. The Real Estate MCP server pulls all three through one endpoint, normalizes price/sqft/yield-implied-cap-rate, and exports a single Excel comparing cap rates by metro. Total monthly Apify spend: ~$80-150 for 5,000 records across all three geographies.
Proptech engineer: comp-set engine
Most proptech apps need an automated valuation model (AVM) underneath. Building an AVM requires (a) recent comparable sales, (b) active listings for asking-price benchmarks, (c) Zestimate or Redfin estimate as a third-party signal. The MCP server returns all three with normalized address geocoding, so the AVM engineer can spend their time on the model rather than on scrapers. Pair with the Google Maps Scraper for nearby-amenity features.
Indie hacker: build a Zillow alternative for a niche
Vertical real-estate sites (luxury, vacation rentals, mid-century homes, off-market) typically lack the inventory budget to license Zillow’s IDX feed. The Real Estate MCP server lets indie hackers spin up a vertical-niche real-estate search experience with pay-per-record economics — meaning you only pay for the listings you actually display or that your users actually query.
Singapore HDB investor: resale-price tracking
Singapore’s HDB resale market has its own dataset published by HDB itself, but the raw download is slow, unjoined, and lacks block-level aggregation. The MCP server wraps the SG HDB Resale Prices actor with sensible filters (town, flat-type, lease-remaining), and the SG URA Property Transactions actor for private property — both delivered to your AI agent or to a CSV.
Sister actors — compose the stack
The Real Estate MCP server bundles these underlying actors. Call any one directly via standard Apify API, or via MCP for AI-agent workflows.
- Redfin Real Estate Scraper — US listings, Redfin estimates, comparable sales
- Rightmove UK Real Estate Scraper — UK active listings + sold-history by postcode
- SG HDB Resale Prices — Singapore HDB resale-flat official transactions
- SG URA Property Transactions — Singapore URA private-property transactions
- Boliga Denmark Real Estate — Denmark residential listings + sale history
- Real Estate MCP Server — unified MCP endpoint over all six sources
- Finance MCP Server — REIT screening + mortgage-rate data for yield modeling
- Google Maps Scraper — nearby amenities, schools, transit, businesses
Frequently asked questions
What is a Model Context Protocol (MCP) server?
MCP is Anthropic’s open protocol for letting AI agents call tools and access data sources in a standardized way. An MCP server exposes one or more tools (search_listings, get_comparables, get_zestimate, etc.) that Claude Desktop, Cursor, Claude Code, Replit Agents, and other MCP-aware clients can invoke directly. You add the server URL and token to your client config once, and from then on the agent has live access to real-estate data. No glue code, no REST wrappers, no scraping logic in your prompt.
How does this compare to HouseCanary or ATTOM Data?
HouseCanary and ATTOM are enterprise data vendors — they require annual contracts, procurement cycles of 4-8 weeks, minimum spend in the $10K-50K/yr range, and US-only coverage. The Real Estate MCP server is pay-per-result on Apify with a free starter tier, immediate signup, and US + UK + Singapore + Denmark coverage out of the box. For teams that need batch AVM scoring on a fixed portfolio at scale, HouseCanary’s AVM may still be the better fit; for everyone building AI agents, prototypes, or vertical-niche real-estate apps, the MCP server is faster and cheaper.
Is Zillow Zestimate access still available?
Zillow officially deprecated their public Zestimate API for new developers years ago — the program is closed unless you have a pre-existing partnership. The Real Estate MCP server retrieves Zestimates directly from public Zillow property pages, which is legal for public data but is rate-limited by Zillow’s anti-bot infrastructure. We handle the rate limiting, retries, and rotation transparently. For very high-volume Zestimate access, pair this with batched scheduling so you stay under Zillow’s per-IP thresholds.
What’s the typical monthly cost for an AI agent app?
An indie hacker shipping a real-estate AI agent app to 100-500 users typically spends $40-120/month on the MCP server. A proptech startup running 5,000-20,000 listing queries per month spends $120-500. A family office or analytical team running 50K+ records spends $1,200-3,000. In every tier, that’s 70-95% cheaper than the equivalent HouseCanary or ATTOM contract, and an order of magnitude cheaper than RealEstateAPI’s $299-2,999/month tiers for similar volume.
Does the MCP server work with Cursor and Claude Code?
Yes. The server speaks standard MCP and works with any MCP-aware client: Claude Desktop, Cursor, Claude Code, Replit Agents, Continue, Cline, and custom clients built with the MCP TypeScript or Python SDK. Add one entry to your mcp.json or client config with the server URL + your Apify token, restart the client, and the real-estate tools appear in your agent’s tool list.
What about data freshness?
All scrapers run on-demand against the live source. Zillow and Redfin pages update within minutes of listing changes; Rightmove updates within an hour. Singapore HDB and URA publish their transaction datasets weekly — the MCP server reflects the most recent published dataset. Boliga is real-time. You can also schedule pre-warmed runs (e.g., a nightly sweep of Austin listings) to keep your own cache fresh.
Is this legal? Zillow / Redfin / Rightmove terms?
All scraped data is publicly available — no login walls, no paywalls, no rate-limit circumvention beyond standard anti-bot fingerprint randomization that all commercial scrapers use. Singapore HDB and URA data is officially published open data. Rightmove and Boliga listings are public marketing content. We do not bypass authentication or scrape user-private information. As with any scraped data, the responsibility for downstream use (resale, redistribution) lies with you — we provide the access layer, you handle compliance with the source terms for your specific use case.
How do I get started?
Sign up for a free Apify account at apify.com/nexgendata — $5/month in free credits is enough to test 50-200 queries. Add the Real Estate MCP server URL to your Claude Desktop, Cursor, or Claude Code config with your Apify token, and your AI agent has live real-estate tools within 2 minutes. From there, layer in the per-source scrapers (Redfin, Rightmove, HDB, URA, Boliga) directly via the Apify API for batch jobs, or keep using MCP for agent-driven workflows.
NexGenData publishes 260+ buyer-intent actors covering real-estate, finance, lead-generation, news + press releases, and competitive intelligence — all pay-per-result. Browse the full catalog →